Single Precision Float Matrix Multiplication (gemm) Performance Benchmark

Device:
Operation System:
Data Source:
Raw Results:
Compute: C = alpha * A * B + beta * C, alpha = , beta =
Max Dimensions: M= N= K=, step down:
Data Layout: A[M][K], B[K][N], C[M][N], Column Major
Date:
Time Unit: milliseconds
Description: each row contains time cost of "times" column times matrix multiplication. Δ column is the variance of each type compared with base type.
No.MNK